Previous Page Table of Contents Next Page

Dietary changes and the health transition in South Africa: implications for health policy

N.P. Steyn, D. Bradshaw, R. Norman, J.D. Joubert, M. Schneider and K. Steyn


South Africa is a middle-income country with a variety of living conditions ranging through wealthy and middle-income suburbs, deprived peri-urban areas, rural farms and undeveloped rural areas. Changing social, political and economic factors have resulted in increased urbanization and changes in diet and health behaviours. Estimates for South Africa show that despite the high burden of infectious diseases, non-communicable diseases (NCDs) account for a large proportion of deaths. In 2000, infectious diseases accounted for 44 percent of deaths, and HIV/AIDS alone for 29 percent (Bradshaw et al., 2003). NCDs accounted for 37 percent of deaths; cardiovascular disease (CVD) and diabetes accounting for 19 percent, and cancers for a further 7.5 percent. In contrast, nutritional deficiencies related to undernutrition accounted for 1.2 percent of deaths. As a result of the relatively high burdens of injuries and HIV/AIDS, the burden of disease in South Africa has been described as a "quadruple burden" of conditions related to underdevelopment, emerging chronic diseases related to unhealthy lifestyles, HIV/AIDS, and injuries.

This case study provides data from published research of diet, dietary trends, nutritional status and diet-related chronic diseases in South Africa over recent decades. These are assessed in the context of trends in the communicable disease burden. A review of the changes in diet and the health transition experienced in South Africa could contribute to the development of a national strategy with a strong dietary policy component that would be effective in the long term.

Brief historical background

South Africa has a heterogeneous population of approximately 46 million people of diverse origins. Historically, people of Khoi, San, Bantu, European and Indian descent pioneered the country, and at present more than a million people are from other African countries, Asia, Europe, Australia, New Zealand and the Americas. The rich heritage of South Africa has resulted in vast cultural and ethnic diversity, with 11 official languages and several other indigenous languages and dialects. The largest organized religion is Christianity, and others include Islam, Hinduism and Judaism. In addition, many people follow a "traditionalist" belief system (Department of Health, SAMRC and Measure DHS+, 2002). The 2001 census (Statistics South Africa, 2003) incorporated the following self-classified and self-reported population groups: black/African (79 percent), coloured/mixed origin (8.9 percent), white (9.6 percent) and Asian/Indian (2.5 percent).[11]

Segregation and discrimination had been part of South Africa’s history for hundreds of years. Over the last century, the country’s people endured complex systems of neo-colonial and Apartheid repression and oppression. In the 1980s, escalating conflict, civil unrest, changes in the ideology of the then-ruling National Party, declining economic growth and international sanctions contributed to the creation of alternative political views. Subsequent negotiations resulted in the country’s first democratic elections in April 1994 and the development of a new political dispensation (Blaauw and Gilson, 2001). South Africa is currently undergoing a profound social transition from its segregationist past to a democracy supported by a progressive constitution entrenching extensive human rights and fundamental political freedoms. The country’s political past was intertwined with its geographic formation and governing system. Hence, 11 geopolitical areas consisting of the former provinces, four independent states and six self-governing areas have been restructured into nine provinces (Figure 1), and a new governing system has been established at the national, provincial and local levels (Blaauw and Gilson, 2001).

The development challenge faced by South Africa is enormous. While aiming to build a society based on human rights and social justice, the country has to grapple with the legacy of an income distribution that is among the most unequal in the world, combined with high levels of poverty and unemployment. Furthermore, strategies to promote economic growth are likely to reduce the likelihood of eliminating these inequalities in the near future (Terreblanche, 2004).

Demographic and socio-economic indicators

The dietary and individual risk behaviour of people defines their nutritional status, health, growth and development. These do not occur in a vacuum but within a cultural, economic, social and political context, which can either aggravate or promote health (WHO, 2003). As South Africa is undergoing major transformations, it is important to describe some demographic, economic, health and development trends that may play diverse roles in nutrition and health.

Selected demographic indicators

Common to the situation in most sub-Saharan countries, South Africa’s demographic and epidemiological data systems have limitations. Determined efforts over the past decade have improved the processes and products of vital registration systems, but sources of complete and reliable vital statistics remain difficult to achieve (Bradshaw et al., 2003). The country’s rapidly growing AIDS epidemic has affected many demographic and epidemiological trends in atypical ways that would challenge data systems under even optimal circumstances. The internationally acknowledged model of the Actuarial Society of South Africa (ASSA) has the best potential for the purposes of this case study, which has used the ASSA2002 suite of models when empirical data were not available or reliable. Details of the models and their assumptions are available on the Internet at

Current geographic composition of South Africa

Source: HSRC, 2003.

Life expectancy and adult, child and infant mortality rates: After a steadily increasing average life expectancy at birth throughout the 1980s, the mortality impact of the country’s severe AIDS epidemic is evident in the considerable drop in life expectancy in the early 1990s, from 61.6 years in 1992 to 49.7 in 2006, and is also reflected in increased infant and child mortality. The harshness of the impact on women is made clear by the unusually rapid narrowing of the difference between female and male life expectancy, from eight years in the early 1990s to less than four years in 2010 (Figure 2). Both the fall in life expectancy and the change in the sex differential are mirrored in the steeply upward trend of the 45q15, or the probability that a person aged 15 years will not live another 45 years to reach 60 years of age (Figure 3). The infant and child mortality rates do not reflect the country’s middle-income economic status, particularly since the AIDS epidemic. However, assuming dedicated efforts to prevent vertical HIV transmission in the model, a recovery to and improvement of pre-AIDS trends in infant and child mortality are projected (Figure 4).

Average life expectancy at birth, 1985 to 2010

Source: ASSA2002 (ASSA, 2004).

Adult mortality,1 1985 to 2010

1 nqx represents the proportion of people in a particular cohort who are alive at the beginning of an indicated age interval (x) and who will die before reaching the end of that age interval (x + n). In other words, the nqx values stand for the probability that a person at his/her xth birthday will die before reaching his/her x + nth birthday. So, 45q15 represents the probability that people aged 15 years will die before they reach the age of 60 (or 15 + 45). This can also be called "premature adult mortality". The 45q15 is widely used as a demographic indicator of adult mortality.

Source: ASSA2002 (ASSA, 2004).

Infant and child mortality rates, 1985 to 2010

Source: ASSA2002 (ASSA, 2004).

Total population and fertility rates: South Africa’s total fertility rate has been declining for several decades, and is currently estimated at 2.6 children born alive per woman during her reproductive lifetime, indicating that the population is well-advanced in its fertility transition (Moultrie and Timæus, 2003). With increasing mortality rates and decreasing fertility and birth rates, the average annual growth rate of the total population is projected to decline dramatically over a short period. These demographic changes reflect an epidemic with a vast impact. Figures 29 and 30, presented in another section of this case study, convey part of the vastness by illustrating the projected numbers of HIV-infected and AIDS-sick people, and showing the huge mortality from AIDS alone compared with that from all other diseases, disabilities and injuries combined.

Urbanization trends: Urbanization and other migration patterns are perceived as important issues in health and nutrition, but relationships and patterns of migration are complex in South Africa, and suitable data sources are very scarce. The 1996 population census provides data on internal migration for the entire population for the first time, but the absence of suitable data prior to this has constrained the analysis of migration data over time (Kok et al., 2003). Urbanization, in particular, has different histories for the country’s four main population groups, with the urbanization levels of black South Africans diverging most prominently from those of other groups. Until July 1986 when it was abolished, "influx control" legislation prevented the black population from settling permanently outside the independent and self-governing states. The "group areas" legislation, repealed in June 1990, enforced the resettlement of millions of South Africans, mostly black African people (Gelderblom and Kok, 1994; Kok et al., 2003). These and other political controls and legislation were directed not only towards restricting black migration, but also towards controlling black labour (Terreblanche and Nattrass, 1990). By 2001, almost 90 percent of the white and coloured population and nearly 100 percent of the Indian population were urbanized, compared with about 50 percent of the black population (P. Kok, personal communication, 2004).

Selected socio-economic indicators

Gross domestic product (GDP): Sufficient, safe and varied food supply can prevent under- and overnutrition and reduce the risk of chronic disease. However, there is also evidence that poverty and inequity are part of the root causes of malnutrition (WHO, 2003). South Africa’s per capita GDP, corrected for purchasing power parity (PPP) at US$11 240 per year in 2001, placed it among the 50 wealthiest nations in the world (May, 2004). However, in 1993 the World Bank described the country as one of the world’s most unequal economies, with a Gini coefficient for income as high as 0.58 (World Bank, cited in May, 2004); this indicator had deteriorated to 0.69 in 2000, making South Africa the third most unequal society in the world (UNDP, 2001). This also suggests that income inequality has worsened nationally, despite official efforts to increase wages at the lower end of the income scale, such as for domestic and farm workers (cf. Department of Labour, 2002).

Figure 5 shows the country’s per capita GDP, corrected for PPP, which has increased steadily since the 1980s. However, such macroeconomic indicators conceal important concerns that may affect community or individual nutritional status and well-being. For example, in 1993, 19 million people - almost half of the country’s population - were categorized as poor (Klasen, 1997 cited in May, 1998), and 11.5 percent of the population were living on less than PPP$1 per day, while 35.8 percent lived on less than PPP$2 (World Bank, 2000 cited in May, 2004). In a rigorous analysis of poverty and related data, Woolard and Leibbrandt (2001 cited in May, 2004) used 1995 data to indicate that the situation continues to be bleak, with 40 to 50 percent of South Africans categorized as poor, including 25 percent ultra-poor. Although definitions of poverty have been adapted over the years and changes in the incidence and severity of poverty are debated, various studies suggest that poverty levels and the number of people living in poverty have increased over recent years (cf. Budlender, 2000; Statistics South Africa, 2002; Van der Ruit and May, 2003; Meth and Dias, 2004 - all cited in May, 2004).

GDP per capita: 1980, 1990 and annual values for 1994 to 2002

Data points for 1981 to 1989 and 1991 to 1993 are interpolated.
Source: Quatec Dataset in UNDP, 2003.

Unemployment: Differing conceptual, methodological, theoretical and ideological positions can influence the measuring of unemployment, and this seems to be particularly true in South Africa (Archer et al., 1990). However, there is wide consensus that the unemployment rate has increased considerably over the past three decades (Figure 6). Since the mid-1970s, every year there have been fewer wage jobs available than the number of people entering the labour market (Archer et al., 1990). Towards the turn of the century this observation was highlighted by May (1998), who said that the South African economy is creating employment too slowly to make a meaningful impression on unemployment levels. Despite employment creation efforts by the new government, after ten years of democracy the rate of unemployment had risen significantly - whether unemployment be defined broadly or narrowly (HSRC, 2003).

Unemployment rates, 1970 to 2002

Sources: South African Reserve Bank Quarterly Bulletin 2003, q1; EIU Country Data; World Bank Global Development Indicators; and IMF Financial Statistics - reported in UNDP, 2003.

Housing and sanitation: The environment that people live in has the potential to aggravate or promote their health. Despite improvements over the past decade, almost a third of South Africa’s households live in informal and traditional dwellings, about a third have piped water inside their homes, slightly more than half use a flush or chemical toilet, and 14 percent have no toilet. A considerable number of households continue to lack basic services, and much still has to be done to enhance the country’s inherited skewed system of access to these services.

Burden of disease

The initial burden of disease study of 2000 (Bradshaw et al., 2003) provides the first set of estimates of the causes of mortality experienced in South Africa. This study made use of several sources of cause of death data, together with the ASSA model to overcome the underregistration of deaths and the misclassification of causes. Figure 7 shows the age distribution of the estimated number of deaths in 2000 by broad cause group. The distinct age pattern of AIDS deaths among children and young adults is clear. Communicable diseases occur across all ages, while injuries affect particularly young adult men. NCDs occur in adult age groups. More such deaths occur under the age of 60, reflecting the age structure of the population. The South African National Burden of Disease Study (SANBDS) estimates that in 2000 NCDs accounted for 37 percent of deaths, followed by HIV/AIDS, which accounted for 30 percent. NCDs accounted for 40 percent of female and 36 percent of male deaths. Stroke is the most common fatal NCD among women, and ischaemic heart disease (IHD) among men. Hypertensive heart disease, diabetes mellitus and chronic obstructive pulmonary disease were also among the leading causes of fatal NCDs in 2000. These conditions coexist with low birth weight, protein-energy malnutrition and other infectious diseases as leading causes of death.

Male and female deaths by age and cause, 2000

Source: 2000 SANBD in Bradshaw et al., 2003.

Dietary trends and associated risk factors

Changes in total dietary energy, carbohydrate, protein and fat intakes

The food balance sheets for 1962, 1972, 1982, 1992 and 2001 were used to describe trends in per capita consumption and are presented in Annex 1 (FAO, 2004). The contributions of different macronutrients to total energy intake are shown in Figure 8. These ratios have not changed much, even though the available per capita energy supply has increased by more than 300 kcal. It is important to remember that food balance sheets present total amounts of food available (not consumed) and do not account for how commodities are distributed according to region, socio-economic sector, gender or other demographic factor. These data are regarded as very crude estimates of dietary intake and have only been included because national data on dietary intake surveys are not available prior to 1999.

Trends in dietary energy supplies from fat, protein and carbohydrate (CHO), 1962 and 2001

Source: FAO, 2004.

However, certain trends emerge for the 40-year period (Annex 1). The per capita available energy supply increased from 2 603 kcal/day in 1962 to 2 921 kcal in 2001, available protein supplies increased from 68.4 to 75.1 g, fat from 61.2 to 79 g, and available carbohydrate supplies from 445 to 478 g. The implication is that at the national level more food is available to consumers. However, the increase in fat availability per capita may be detrimental to health from a chronic diseases perspective.

The first nationally representative dietary study in South Africa - the National Food Consumption Survey (NFCS) - was undertaken in 1999 (Labadarios et al., 2000). It was a cross-sectional survey of children aged one to nine years, with provincial representation drawing on the 1996 census data. The aim of the survey was to collect baseline data from which to formulate appropriate policy guidelines for food fortification, as well as to develop appropriate nutrition education material for South African children. The final sample comprised 2 894 children, with a response rate of 93 percent. Socio-demographic, dietary and anthropometric data were collected for each participant.

As the NFCS was the first national dietary study in South Africa, it is not possible to compare macronutrients over time in a reliable manner. However, by examining two studies, one in adults and one in schoolchildren, some changes can be deduced. Bourne (1996) examined the macronutrient intake of black adults living in Cape Town (Figure 9). Certain trends are noticeable, for example, the intake of carbohydrate calculated as percentage of total energy intake decreased from 61.4 to 52.8 percent as people spent more of their lives in the city. In contrast, fat intake increased from 23.8 to 31.8 percent according to time spent in the city.[12] Protein intake remained more or less the same over time, although the contribution of animal protein increased, whereas the amount of plant protein decreased. Fibre intake (not shown), also decreased significantly - from 20.7 to 16.7 g - with increased time living in the city.

These changes are all consistent with a population undergoing the nutrition transition, i.e., changes in diet from a traditional high-carbohydrate, high-fibre, low-fat diet to one with higher fat and sugar intakes and lower carbohydrate and fibre intakes (Popkin, 2001).

Changes in contributions of macronutrients to total energy intake among black adults (19 to 44 years) according to percentage of time lived in the city (Cape Town)

Sample size: 649.
Source: Bourne, 1996.

In urban areas of Gauteng, mean fat intake increased from 17 percent in 1962 to 25.8 percent in 1999, while carbohydrate intake decreased from 72 to 60.3 percent, as shown in Figure 10 (Lubbe, 1973; Labadarios et al., 2000). Some differences between the two studies need to be kept in mind, however. Results from the 1962 study are reported for six- to nine-year-olds, using a modified diet history, while results from 1999 are for one- to nine-year-olds, using a 24-hour recall dietary method. Despite these differences, schoolchildren showed similar patterns of macronutrient intakes to those of adults in the Cape Town study by Bourne (1996). These two studies support the trends that energy and fat intakes have increased since 1962, as shown by the food balance sheets.

Macronutrient distribution as percentages of total energy consumed by black schoolchildren in urban areas of Gauteng in 1962 (six to nine years) and 1999 (one to nine years)

Sample sizes: 1962, 552; 1999, 427.
Sources: 1962 - Lubbe, 1973; 1999 - NFCS, 1999 (Labadarios et al., 2000).

Differences in nutrient intake among ethnic groups and between urban and rural areas

It is important to note that there is a diversity of ethnic and cultural groups in South Africa with different traditional eating patterns. The white population consume a typical Western diet, which has a high fat intake (> 30 percent of total energy), low carbohydrate intake (< 55 percent energy), low fibre and high free sugar intake (> 10 percent energy) (Wolmarans et al., 1989). The Indian and coloured (mixed ancestry) populations have a very similar pattern to this, albeit each group consumes certain foods more commonly (Langenhoven, Steyn and van Eck, 1988). The black African population has two distinct types of eating patterns. The rural population still follows a very traditional diet, which is high in carbohydrates (> 65 percent of total energy), low in fat (< 25 percent total energy), low in sugar (< 10 percent total energy) and moderately high in fibre (Steyn et al., 2001). On the other hand, the urban black African population demonstrates an adoption of the Western diet of the other groups. The carbohydrate (< 65 percent total energy) and fibre intakes of this group are lower, and its fat intake is higher (> 25 percent total energy) (Bourne et al., 1993).

In Figure 11, macronutrient distributions show marked differences among whites, urban blacks and rural blacks. White males (aged 35 to 44 years) have the highest intakes of fat, protein and added sugar and the lowest intake of carbohydrates (Wolmarans et al., 1988). Rural black adults of the same age have the highest intake of carbohydrates and the lowest of protein, fat and added sugar (Steyn et al., 2001). Black urban males lie between the two extremes (Bourne et al., 1993). This figure suggests that there is a diet transition among urban blacks from a traditional rural diet to an urban one that is approaching the completely Westernized diet of the white population. However, it should be remembered that the studies were not undertaken at the same point in time, which may have influenced the results.

Macronutrient distribution as percentages of total daily energy among adult white urban (15 to 64 years), black urban (19 to 44 years) and black rural (20 to 65 years) males

Sample sizes: white urban, 454; black urban, 285; black rural, 74.

Sources: white - Wolmarans et al., 1989; black urban - Bourne et al., 1993; black rural - Steyn et al., 2001.

Table 1 shows the differences in dietary intake among all the main ethnic groups (males) in South Africa, with urban and rural subgroups for blacks (Bourne et al., 1993; Langenhoven, Steyn and van Eck, 1988; Steyn et al., 2001; Vorster et al., 1995; Wolmarans et al., 1988; 1999). These studies were geographically and ethnically representative of the areas where they were undertaken, and can be regarded as a good reflection of the typical diet of each specific group. The white, Indian and coloured groups have the highest intakes of fats, protein and free sugar, which are not in line with the WHO/FAO (2003) recommendations. Black males in rural areas have the lowest intakes of all types of fat and protein. Urban males, once again, illustrate the nutrition transition that has taken place. Table 2 shows various transitions that have taken place in the black population (MacIntyre et al., 2002).

The urban upper income group has the highest fat and protein intakes as a percentage of energy intake. This group also has the highest cholesterol intake, which is higher than the WHO/FAO (2003) recommendation (< 300 mg/day). At the other end of the scale are rural residents and rural farm workers, who have a prudent diet that is low in fat and high in carbohydrate.

Comparison of macronutrient mean ranges in six dietary studies in adult males and females

Dietary factor

white rural
n = 1 113
15-64 years

black rural
n = 210
20-65 years

white urban
n = 317
15-64 years

black urban
n = 983
19-44 years

Indians urban
n = 370
15-69 years

coloured urban
n = 276
20-34 years

WHO goals
% energy

Energy (kJ) 1







Energy (kcal)

1 500-3 024

1 434-1 590

1 405-2 976

1 386-2 035

1 190-2 024

1 690-2 452

Total fat (% E) 2








SFA (% E) 3







< 10%

PUFAs (% E) 4








CHO (% E) 5








Free sugar (% E)







< 10%

Protein (% E)








Cholesterol (mg)



140-176 mg/4.2kJ


76-117 mg/4.2kJ


£ 300 mg/day

1 kJ = kilojoules.
2 E = energy.
3 SFA = saturated fats.
4 PUFA = poly-unsaturated fats.
5 CHO = carbohydrate.
Sources: CORIS - Wolmarans et al., 1988; DIKGALE - Steyn et al., 2001; VIGHOR - Vorster et al., 1995; BRISK - Bourne et al., 1993; Indian Study - Wolmarans et al., 1999; CRISIC - Langenhoven, Steyn and van Eck, 1988; WHO/FAO, 2003.

Distribution of macronutrients in the diet of black South African males (15 to 80 years), by area and income

Dietary factor

n = 194

Farm workers (rural) low-income
n = 109

Urban (informal settlement) low-income
n = 128

Urban middle-income
n = 229

Urban high-income
n = 83

WHO7 goals
% of energy

Energy (kJ)






Energy (kcal)

2 285

2 122

2 222

2 356

2 338

Total fat (% E)






15 - 30%

CHO (% E)






55 - 75%

Protein (% E)






10 - 15%

Cholesterol (mg)






£ 300 mg/day

Fibre (g)






Sources: MacIntyre et al., 2002; WHO/FAO, 2003.

Changes in intakes of types of food and food groups over time

According to FAO data (Annex 1), cereal availability increased from 169.3 kg per capita per annum in 1962 to 187.8 kg in 2001; as did the availability of starchy roots (from 13 to 29.7 kg), vegetable oils (5.7 to 14.5 kg), fruits (24.1 to 36.0 kg), alcohol (43.8 to 56.8 litres), meat (31.6 to 37.5 kg), eggs (2.5 to 6.1 kg) and fish (5.5 to 7.9 kg). Foods whose per capita availability decreased were sugar and sweeteners (from 39.4 to 32.8 kg), offal (4.5 to 3.8 kg), animal fats (including butter) (3.0 to 0.7 kg) and milk (78.0 to 54.1 kg).

These data reflect a number of scenarios. Availability of staple cereals gradually increased, as did that of other items mentioned in the previous paragraph; these increases account for the overall increase in energy intake. Vegetable oil and meat per capita also increased significantly, which accounts for the large increase in fat and saturated fat intakes. Of concern is the fact that vegetable availability remained constant (at 43.5 to 44.2 kg per annum). Overall fruit and vegetable availability was 185 g/day (excluding starchy roots), which is far less than the recommended intake of 400 g/day (WHO/FAO, 2003). This has serious implications, because low fruit and vegetable intake is a risk factor for many NCDs.

Among black adults, the amounts consumed from different food groups changed with increasing time spent in Cape Town, as shown in Figure 12. Groups for which consumption rose were meat, fruit and vegetables, fats and non-basic foods (such as drinks and sweets), while consumption from the dairy and cereal groups decreased. Similar findings are presented for young females in Figure 13. The higher consumption of sugar-containing food items in urban compared with rural areas is shown in Figure 14.

Percentage contributions of different food groups to energy, by time spent in Cape Town by black adults (19 to 44 years)

Sample size: 649.
Source: Bourne, 1996.

Food consumption of black female university students from urban and rural areas

Sample size: 115.
Source: Steyn et al., 2000

Percentage consumption of sugar-containing items by children aged six to nine years

Sample size: 439.
Cordial = a drink made from sweet sucrose concentrate.
Cold drink = soft drink.
Source: NFCS in Labadarios et al., 2000.

Current diet

The results from NFCS provide the first nationally representative dietary data for South Africa. Table 3 indicates the mean nutrient intakes of children and compares them with recommended nutrient intakes (RNI) (FAO/WHO, 2002). Overall, the energy intakes of both rural groups were less than the RNIs, as were the intakes of vitamins A and C, niacin, vitamin B6 and zinc. For folate and calcium, urban and rural intakes were less than the RNIs. An important aspect of the study was the disparities in intakes between urban and rural areas. For most nutrients, the mean values in urban areas were significantly higher than those in rural ones.

To understand the dynamics of dietary change, the main food groups consumed by South African adults and children in urban and rural areas were examined (Table 4). In lieu of the lacking national data on adults, data from combined databases were summarized using secondary data analyses to show the dietary intakes of adults (Steyn et al., 2001; Nel and Steyn, 2002) and children (aged one to five years) (Labadarios et al., 2000). Although rural dwellers have higher cereal and vegetable intakes, urban adults and children far exceed the rural people’s consumption of most other food groups, particularly sugar, meat, vegetable oil, dairy, fruit, roots and tubers and alcohol.

Mean nutrient intakes of children


Children 1-3 years (n = 1 308)

Children 4-6 years (n = 1 083)







#Energy (kJ)

4 403 (2 043)

3 992* (1 790)

4 200 (1 933)

5 614 (2 375)

4 963* (2 283)

5 271* (2 349)

Energy (kcal)

1 048 (486)

950* (426)

1 000 (460)

1 337 (565)

1 182* (544)

1 255* (559)

CHO (g)

154 (72)

151 (71)

152 (72)

192 (80)

193 (91)

193 (86)

#Added sugar (g)

26 (23)

18 (17)

22 (21)

36 (30)

24 (34)

29 (33)

#Protein (g)

33 (18)

29 (17)

31 (18)

43 (21)

36 (19)

39 (21)

Fat (g)

29 (21)

22 (16)

25 (19)

38 (25)

42 (21)

31 (24)

Fibre (g)

9 (6)

10 (7)

9 (6)

13 (7)

13 (8)

13 (8)

#Vitamin A (RE)

463 (943)

252* (349)

359* (723)

544 (1 313)

319* (1 007)

425* (1 167)

#Vitamin C (mg)

41 (96)

20* (36)

31* (73)

36* (65)

29* (78)

33* (72)

Thiamine (mg)

0.6 (0.3)

0.6 (0.3)

0.6 (0.3)

0.7 (0.4)

0.7 (0.4)

0.7 (0.4)

#Riboflavin (mg)

0.8 (0.8)

0.6 (0.6)

0.7 (0.7)

1.0 (1.0)

0.7 (1.0)

0.8 (0.9)

#Niacin (mg)

6.4 (4.7)

4.8* (3.8)

5.6 (4.3)

9 (6.2)

6.3 * (4.4)

7.6 * (5.5)

#Vitamin B6 (mg)

0.6 (0.4)

0.4* (0.3)

0.5 (0.4)

0.8 (0.6)

0.5 * (0.4)

0.6 (0.5)

#Vitamin B 12 (ug)

2.7 (8.4)

1.4 (4.4)

2.1 (6.8)

3.7 (12.1)

2 (10.2)

2.8 (11.2)

#Folate (mg)

102* (81)

86* (84)

94* (83)

161* (119)

127* (111)

143* (116)

#Calcium (mg)

345* (326)

302* (326)

324* (327)

342* (282)

270 * (254)

304* (269)

#Iron (mg)

4.9* (3.6)

4.7* (3.8)

4.8 * (3.7)

6.7 (4.2)

6.1 (4.6)

6.4 (4.5)

#Zinc (mg)

4.5 (2.7)

3.9* (2.5)

4.2 (2.6)

5.9 (3.3)

4.8* (3.1)

5.3 (3.2)

* = mean intake is less than the FAO/WHO (2002) RNI.
# = significant urban/rural differences (p < 0.01).
Source: Steyn et al. in Labadarios et al., 2000

Adults’ and children’s consumption, by food group and residence

Food group

Adults and children 10+ years (n = 817)

Children 1-5 years (n = 2 048)

RSA g/day

Urban g/day

Rural g/day

RSA g/day

Urban g/day

Rural g/day




1 023











Stimulants: tea, coffee














Meat and offal







Vegetable oils










































Roots and tubers







Nuts and oilseeds




























Animal fat







Source: Nel and Steyn, 2002.

Changes in alcohol intake

According to the food balance sheets, per capita alcohol consumption in South Africa increased between 1962 and 2001 (Annex 1). Certain trends are noticeable from two surveys carried out in the 1990s: SADHS in adults (Department of Health, SAMRC and Measure DHS+, 2002) and the Youth Risk Behaviour Study (YRBS) (Reddy et al., 2003) in teenagers (Figure 15). YRBS found that more than 30 percent of teenagers had drunk and/or binged on alcohol in the preceding 30 days. According to SADHS, nearly 30 percent of adult males reported using alcohol excessively, based on the CAGE test (Ewing, 1984), compared with 10 percent of females. High alcohol consumption is a risk factor for chronic diseases such as stroke, diabetes and cancer of the oesophagus, liver and breast, and so needs to be addressed as an underlying determinant in the prevention of NCDs (WHO/FAO, 2003).

Prevalence of teenager (13 to 19 years) school attendees reporting drinking alcohol or bingeing in the past month, and prevalence of alcohol-dependent adults

* Consumed an alcoholic drink on one or more days during the previous month.
** Consumed five or more alcoholic drinks on one or more days during the previous month.
+ According to the CAGE questionnaire.
Sources: YRBS; SADHS.

Trends in nutritional status

Trends in the prevalence of undernutrition and protein-energy malnutrition

Nationally representative and comparable anthropometric data over time are only available for 1994 and 1999 in children and for 1998 in adults; hence, they do not show long-term trends. In order to obtain longer time trends, smaller localized studies have been used to provide comparisons with the 1994 and 1999 data on children. Data on one localized and two national studies undertaken in South Africa between 1986 and 1999 are shown in Table 5. Given the differences in children’s ages, conclusions on trends should be interpreted cautiously. The 1986 study sampled black preschool children on farms in areas other than the "homelands", where the greater part of the black population lived. Consequently, the data are not a true reflection of the actual prevalence of malnutrition, which would have been higher if these areas had been included. Before democratization, the health care services provided to the population in "homelands" were totally insufficient. In 1994, the South African Vitamin A Consultative Group (SAVACG, 1995) undertook a national study of preschool children, and the 1999 NFCS included school-going children. These studies showed similar results, with underweight ranging from 6.9 to 10.7 percent, stunting from 16.1 to 27 percent and wasting from 1.8 to 3.7 percent. Malnutrition prevalence was always higher in rural than urban areas. There appears to be a small improvement in the prevalence of stunting between 1994 and 1999 in these two nationally representative surveys.

Two earlier studies (1969 and 1975) were undertaken as representative studies of the Transvaal, now partly Gauteng (Figure 16). These studies used the Harvard-3rd percentile as an indicator of underweight, while the later studies used the National Center for Health Statistics (NCHS) percentiles, and there are some discrepancies even though the two standards are very similar. There are large decreases in the prevalence of underweight in urban and rural areas until 1994. The smaller increases after 1994 in urban Gauteng are probably because of the large migration into this region following the lifting of migration restrictions.

Prevalence of low weight-for-height, height-for-age and weight-for-age in children, 1986, 1994 and 1999

0-591 months

6-71 months

6-71 months

6-71 months

12-721 months

N = 1 745

n = 4 757

n = 6 062

n = 10 819

n = 2 200

Weight-for-age < -2 SD (NCHS)





8.8 (9.7)2

95% confidence interval

6.8; 9.9

6.0; 7.9

9.6; 11.9

8.5; 10.1

7.6; 10.1

Height-for-age < -2 SD





19.3 (22.1)2

95% confidence interval

19.2; 29.7

14.4; 17.8

24.8; 29.3

21.4; 24.5

17.5; 21.2

Weight-for-height < -2 SD





3.7 (3.7)2

95% confidence interval

1.3; 2.3

1.5; 2.7

2.3; 3.4

2.2; 2.9

3.0; 4.4

1 Six to 71 months category not available.
2 12 to 96 months.
Sources: 1986 - Kustner, 1987; 1994 - SAVACG, 1995; 1999 - Labadarios et al., 2000; Steyn et al., 2005

Prevalence of underweight in black preschool children (under 72 months), 1969, 1975, 1994 and 1999

Sample sizes: 1969, 2 073; 1975, 3 655; 1994, 11 238); 1999, 2 200.
Sources: 1969 and 1975 - Richardson, 1977; 1994 - SAVACG, 1995; 1999 NFCS - Labadarios et al., 2000; Steyn et al., 2005.

Trends in the prevalence of overweight and obesity

The prevalence rates of overweight and obese children at the time of NFCS 1999 are shown in Table 6. There were significant differences between urban and rural areas, among location domains and among age groups. Overweight was highest in formal urban areas and in children aged one to three years. The finding that overweight/obesity was higher in urban areas is an indication that the nutrition transition is under way, and that undernutrition and associated infectious diseases should not be the only health concern among policy-makers. The data show that the prevalence of combined overweight and obesity (17.1 percent) is nearly the same as that for stunting (21.6 percent) (Steyn et al., 2005). Furthermore, stunting was associated with an increased risk (OR = 1.80, CI = 1.48-2.20) of being overweight (BMI ³ 25) (Steyn et al., 2005). This finding suggests that stunting in childhood predisposes to overweight or obesity when sufficient food becomes available. This poses a threat for the emergence of chronic disease risk factors when stunted children become obese adults.

Percentages of children with BMI values ³ 25 and ³ 30 using the Cole et al. (2000) cut-off points

BMI cut-off points

Domain analysis by area of residence*

Domain analysis by urban/rural*

Domain analysis by age group*



Formal urban

Informal urban




1-3 years

4-6 years

7-8 years

n = 108

n = 946

n = 272

n = 874

n = 982

n = 128

n = 795

n = 861

n = 544

n = 2 200

% ³ 30 BMI











Lower 95% CI











Upper 95% CI











% ³ 25 BMI











Lower 95% CI











Upper 95% CI












p = 0.0066


< 0.0001

* Chi-square p-value for testing for associations, using weighted values, between BMI groupings, area of residence, urban/rural and age groups.
CI= confidence interval. SD = standard deviation.
Source: 1999 NFCS in Steyn et al., 2005.

SADHS (1998) was the first nationally representative health survey in adults aged 15 years and over, so in order to examine trends from previous years and compare with these data it is necessary to evaluate earlier studies that were representative of specific ethnic groups. Interpretation of these data should keep these limitations in mind. The earlier studies include a baseline study in 1979 on coronary heart diseases risk factors in white adults in three towns of the Western Cape Province (CORIS) (Jooste et al., 1988), and similar studies in 1982 in the coloured population (CRISIC) (Steyn et al., 1985), the black population in Cape Town (BRISK) (Steyn et al., 1991) and the Indian population (Seedat et al., 1990).

Figures 17 and 18 show the extent of obesity as a problem in men and women in South Africa. In women, the prevalence of obesity has remained high in all studies since 1979, particularly in black women, who show the highest prevalence. In men there appears to be a large increase in obesity in whites when the 1979 study is compared with that of 1998.

The most recent SADHS data on overweight and obesity in adults indicate that obesity increases with age until about 35 years in both men and women and declines from about 55 years. More than 40 percent of women aged 35 years and over are obese, and more than 20 percent of all women are overweight. For the ethnic groups, obesity is highest in black women and white men. At 12.9 percent and 5.6 percent, respectively, for men and women, the prevalence rate of underweight (BMI < 18.5) in adults is far lower than those of overweight and obesity (Department of Health, SAMRC and Measure DHS+, 2002).

The rising prevalence of obesity in South Africa gives cause for serious concern because of the increased risk of diabetes and CVD (WHO/FAO, 2003). These diseases have direct costs, which may be as high as 6.8 percent of total health care costs, as well as indirect costs, such as workdays lost, doctor visits, impaired quality of life and premature mortality (WHO/FAO, 2003).

The data presented in this section of the case study illustrate that both over- and undernutrition exist in South Africa, with the extremes being most clearly seen among the black population.

The prevalence of obesity (BMI ³ 30) in women, 1979 to 1998

Sample sizes: 1979, 3 831; 1982, 498; 1990, 544); 1998, 7 970.
Sources: 1979 - Jooste et al., 1988; 1982 - Steyn et al., 1985; 1990 - Seedat et al., 1990; Steyn et al., 1991; 1998 - Department of Health, SAMRC and Measure DHS+, 2002.

The prevalence of obesity (BMI = 30) in men, 1979 to 1998

Sample sizes: 1979, 3 357; 1982, 478; 1990, 442; 1998, 5 558.
Sources: 1979 - Jooste et al., 1988; 1982 - Steyn et al., 1985; 1990 - Seedat et al., 1990; Steyn et al., 1991; 1998 - Department of Health, SAMRC and Measure DHS+, 2002.

Trends in micronutrient status

The nationally representative SAVACG survey in 1994 examined, among other factors, the vitamin A and iron status of children aged 0 to five years (SAVACG, 1995). This was the first study to examine micronutrients in children at the national level. Micronutrient intakes of children will be measured again in 2005.

About 3 percent (Figure 19) of the sampled children showed serum vitamin A deficiency (VAD) (serum retinol < 10 ug/dl), while 33 percent were marginally deficient (serum retinol < 20 ug/dl) (SAVACG, 1995). Children in the 36 to 47 months age group were the most affected, with 11 percent having haemoglobin concentrations of less than 11 g/dl and 25 percent with low iron stores (ferritin < 12 ug/dl) (Figure 20). The mandatory fortification of maize and wheat with vitamin A, iron and other micronutrients since 2003 is expected to decrease these micronutrient deficiencies in South African children in the future.

Results from an iodine deficiency survey in 1998 in primary schoolchildren show that within provinces between 0 and 42 percent of schools had children who were iodine deficient (Immelman, Towindo and Kalk, 2000) (Figure 21). Schools in rural areas of Mpumalanga and Limpopo provinces were most affected. The survey also found that mandatory salt iodization since 1995 had considerably improved the iodine status of children. However, there are still some minor weaknesses in the national salt iodization programme, such as the use of non-iodized salt in 6.5 percent of households and the under- or non-iodization of a substantial percentage of household salt.

Vitamin A status of children aged six to 71 months, 1994

Sample size: 4 283.
Source: SAVACG, 1995.

Iron status of children aged six to 71 months, 1994

Sample size: 4 494.
Source: SAVACG, 1995.

Percent of schools whose students have low median urinary iodine, by province

Sample size: 179 schools.
Source: Immelman, Towindo and Kalk, 2000.

Micronutrient deficiencies continue to contribute to the burden of mortality in South Africa. Preliminary results of a study to assess the burden attributable to selected nutritional deficiencies estimate that in 2000 nearly 3 000 deaths to diarrhoea in children aged 0 to four years were attributed to VAD. Furthermore, about 200 maternal deaths were attributed to VAD in pregnant women, while more than 3 500 perinatal deaths and about 180 maternal deaths (0.7 percent of total deaths) were attributed to iron deficiency anaemia (IDA) (Nojilana et al., in press).

No national data on biochemical deficiencies among adults are available. However, numerous localized studies have shown high prevalence of iron deficiency in women (Kruger et al., 1994; Dannhauser et al., 1999) and of VAD, particularly in HIV-infected adults (Kennedy-Oji et al., 2001; Visser et al., 2003).

Other chronic diseases and associated lifestyle risk factors

Physical inactivity

There is a paucity of data on the physical activity levels of South Africans, making it difficult to show trends over time. Figure 22 shows current levels of inactivity in South African teenagers from a national survey undertaken in 2002 (Reddy et al., 2003). Overall, coloured girls have the highest levels of inactivity, with nearly 60 percent doing little or nothing. The high levels of inactivity go a long way to explaining the high levels of overweight, obesity and hypertension, particularly in women. Figure 23 reports data on inactivity in adults. With the exception of the black population, the prevalence of inactivity was very high (more than 90 percent), both at work and during leisure time.

Percentages of 13- to 19-year-olds reporting insufficient or no physical activity at work,* 2002

Sample size: 10 100.
* Insufficient or no physical activity means the person did not participate in vigorous or moderate activity that would have been sufficient for a health benefit over the previous seven days.
Source: YRBS, 2002 in Reddy et al., 2003.

Percentages of 15- to 64-year-olds reporting insufficient or no physical activity at work (< 32 300 kJ/wk) and during leisure times (< 8 400 kJ/wk)

Sample sizes: white, 7 188; coloured, 976; black, 986; Indian, 778
Sources: white - CORIS in Rossouw et al., 1983; coloured - CRISIC in Steyn et al., 1985; black - BRISK in Steyn et al., 1991; Indian - Seedat et al., 1990.

All these studies identified physical activity patterns by means of questionnaires. The measurement of physical activity by questionnaires is challenging in large epidemiological studies. Consequently, the patterns shown here must be interpreted with caution, but the overall trends suggest that the Indian and white populations are the most inactive at work and leisure, while the black population is the least inactive.

Tobacco intake

Tobacco consumption patterns in South Africa between 1990 and 2004 illustrate the impact of an aggressive tobacco control policy that was phased in during this period. The policy had two distinct aspects: tobacco control legislation and rapidly increasing excise taxes. Major legislative milestones include the Tobacco Controls Act, which was passed in 1993 and introduced health warnings on cigarette packets and advertisements. The act was amended in 1999 with the banning of all advertisements and prohibitions on smoking in all indoor public areas and on selling tobacco products to minors. In 1994, the government announced the phasing in of an increased excise tax, which added 50 percent to the retail price of tobacco. This resulted in a real increase of 256 percent in the excise tax per pack of cigarettes between 1994 and 2004; the real price of cigarettes increased by 127 percent over the same period. WHO’s Tobacco Framework Convention has been ratified by South Africa and a sufficient number of Member States to require all countries to comply with these laws. South Africa is currently expanding its tobacco control legislation to ensure compliance (Van Walbeek, 2005).

In 1992, Martin, Steyn and Yach reported that 31.5 percent of South Africans smoked. The prevalence rate peaked at 34 percent in 1995, as recorded by Reddy et al. (1998), declining steadily thereafter to reach 24 percent in 2003. The average number of cigarettes per smoker per year decreased from 229 packs in 1993 to 163 in 2003. Africans, males, young adults and poorer people experienced the most rapid decreases in smoking prevalence, while the decrease was less pronounced among whites, females and older and more affluent people.

Despite these positive trends, the prevalence of tobacco smoking is still high, particularly among youth. A recent survey found that the prevalence of cigarette smoking (daily and occasionally) was higher in 14-year-old adolescent males than females (21.5 versus 15.7 percent) (Reddy et al., 2003). At 16 years of age, 30.4 percent of males were already smoking cigarettes, rising to 38 percent in 18-year-old males.

In 1998, SADHS found that more than 39 percent of African, white, coloured and Indian adult males (15 years and over) smoked daily or occasionally, with the lowest prevalence in rural black males (37 percent). In rural areas, only 4 percent of black females smoked daily, compared with 6 percent in urban areas. The overall prevalence of daily smoking for females was lowest in the black population (5 percent) and highest among whites (27 percent) (Department of Health, SAMRC and Measure DHS+, 2002). Because tobacco use is a risk factor for heart disease and lung cancer, which are serious contributors to both morbidity and mortality in South Africa, it calls for preventive measures, particularly among youth. The finding that nearly one-third of 16-year-old males are current smokers is a serious concern. Figure 24 shows the prevalence of smoking in adult males for 1984 and 1998. The highest prevalence rates were in whites and Indians. Despite the finding that smoking prevalence peaked in 1995 (Reddy et al., 1998), it is still considerably higher than it was in 1984.

Prevalence of smoking in adult males by population group, 1984 and 1998

Sources: 1984 - Yach and Townshend, 1988; 1998 - SADHS.


No national surveys on serum cholesterol levels have been conducted. Figure 25 presents the results from four studies undertaken in different ethnic groups in localized settings. White men had the highest mean total cholesterol values, and black men the lowest. Indian and coloured men had similar mean values to those of white men, albeit slightly lower. With the exception of black men, all were found to have mean values above the recommended limit of 5.2 mmol/l. Hence, high serum cholesterol was a strong risk factor for CHD in South African men who are not black. These studies were conducted in the late 1970s and early 1980s, and it is not possible to say what trends and changes have taken place in the cholesterol values of black males since then.

Mean total serum cholesterol values (mmol/l) in adult males, by age and ethnic group

Sources: black - Norman et al., unpublished data, 2005; white - CORIS in Rossouw et al., 1983; coloured - CRISIC in Steyn et al., 1985; Indian - Seedat et al., 1990.


SADHS was the first national survey to measure blood pressure of adults, and its findings are presented in Figure 26. The lowest prevalence of hypertension (BP > 140/90 mmHg) was found in black men (20.2 percent) and the highest in white men (38 percent). Coloured and white women and Indian men also had very high prevalence, at close to 30 percent. Of great concern, however, are the levels of hypertension control, namely control of those who have hypertension (BP > 140/90 mmHg). The highest levels of control were found in white women and Indians, and the lowest in black men and coloured women. Fewer than 10 percent of the latter were found to be controlled. The high levels of hypertension illustrated in Figure 26, together with the high prevalence rates of obesity, tobacco use (shown earlier) and hypercholesterolaemia help to explain the high prevalence of CVD in adults.

Prevalence of hypertension (BP > 140/90 mmHg) in adults, 1998

Sample size: 2 049.
BP controlled = among those with hypertension, BP is < 140/90 mmHg, i.e. controlled.
Source: SADHS

Cardiovascular diseases and diabetes

The poor quality of historical cause of death data makes it very difficult to assess trends in mortality. Bradshaw et al. (1995) calculated age-standardized deaths rates by population group for 1985, based on the deaths reported for the years 1984 to 1986 relative to population estimates for 1985. Rates for blacks were calculated for urban areas only because there was high underregistration of deaths for blacks in rural areas. These are compared with estimated death rates for 2000 (Bradshaw et al., 2004). Figure 27 shows age-standardized mortality rates for IHD, stroke, hypertensive heart disease and diabetes in 1984 to 1986 compared with 2000, by population group. These comparisons must be interpreted carefully as the estimates for 2000 have been adjusted for misclassification and underregistration of deaths, but those for 1984 to 1986 have not. The increase in IHD across all groups is likely to be a result of the adjustment for misclassification of ill-defined cardiac causes in 2000. From Figure 27, it appears that hypertensive disease and stroke increased dramatically in the black population. Diabetes mortality increased in all ethnic groups, but most in the black population. The increased rates in the black population may be an artefact of the adjustment for underregistration of deaths in the 2000 estimate.

IHD was the main cause of mortality among CVDs and diabetes in white males between 1949 and 1985 (Bradshaw et al., 1995). IHD mortality increased from 260 per 100 00 in 1949 to more than 300 per 100 000 between 1964 and 1979. It subsequently decreased from 312 per

100 000 in 1978 to 139 in 1989 (Walker, Adam and Küstner, 1993). IHD was also the main cause of CVD mortality in white females, although the rates were about half those of males.

IHD was the main cause of mortality among CVDs and diabetes in coloured males until 1969, when it was replaced by stroke (Bradshaw et al., 1995). There was a very large increase in mortality from IHD between 1958 and 1969; it then remained stable at about 150 per 100 000 until 1985. Stroke was the major cause of mortality in coloured females. Mortality from diabetes increased fourfold, from ten in 1949 to 40 in 1984.

IHD was the major cause of mortality from CVDs and diabetes in Indian males between 1949 and 1985, followed by stroke and diabetes (Bradshaw et al., 1995). During this period the mortality rates remained fairly constant, except that IHD increased to more than 250 per 100 000 and diabetes to more than 70 in 1985. Stroke was the major cause of mortality in Indian females. There was also a dramatic increase in the mortality rate from diabetes, from about 20 per 100 000 in 1949 to 70 in 1984.

These data seem to suggest that both IHD and stroke have high mortality rates and that hypertensive heart disease and diabetes may have been growing. IHD is particularly high for whites and Indians, while hypertensive heart disease and stroke are highest among the black population group.

Age-standardized mortality rates from hypertensive heart disease, stroke, IHD and diabetes, per 100 000 population, 1985 and 2000

Note: 2000 figures are adjusted for misclassification and underregistration of deaths, but not 1985 figures. 1985 figures for blacks are for urban blacks only.

Sources: 1985 - Bradshaw et al., 1995; 2000 - Bradshaw et al., 2003.


In 2000, the South African National Burden of Disease Study (SANBDS) found that cancers as a group (the malignant neoplasms category) accounted for 41 691 deaths (7.5 percent of all deaths), ranking them the fourth leading cause of death for all people and the second leading cause of death for people aged 60 years and over (Bradshaw et al., 2003). In males, trachea/bronchi/lung (also referred to as lung) cancer accounted for 22.5 percent of all cancer deaths, followed by oesophageal cancer (17.2 percent). Among the top causes of cancer deaths in females were cancer of the cervix (17.9 percent), breast (15.7 percent) and lung (10.9 percent).

Lung cancer

There have been marked increases in mortality rates for lung cancer among males of all population groups, with those among whites increasing almost threefold between 1949 and 1979, while those among coloured males increased even more dramatically. Smaller increases are seen among females. In 1984, 34.5 percent of all white deaths could be attributed to smoking-related diseases, compared with 24.5 percent for Indians, 14.5 percent for coloureds and 3.9 percent for blacks (Yach and Townshend, 1988).

From 1984 to 1986 the age-standardized mortality for lung cancer was highest in coloured urban males (88.4/100 000), followed by white urban males (48.7/100 000), black urban males (27.9/100 000) and Indian urban males (21.8/100 000) (Bradshaw et al., 1995). National age-standardized death rates for 2000 (not disaggregated by urban/rural residence) showed that coloured males had the highest rates (82.1/100 000), followed by white males (54.3/100 000). In 2000, the age-standardized death rates of black males and females were 33.4 and 6.0 per 100 000, respectively (Bradshaw et al., 2003).

The differences in smoking rates among the population groups at different stages of the tobacco epidemic, as well as the gender differences, are reflected in the age-specific death rates for lung cancer. (It is important to note, however, that these death rates reflect exposure to tobacco in the past.) In the older age groups, black men had lower rates than coloured and white men, but the lung cancer death rates in black males in the 35 to 44 years and 45 to 54 years age groups were higher than those in the white population. This, however, is not seen in black women, who have lower lung cancer death rates at all ages.

Oesophageal cancer

In South Africa, the incidence of oesophageal cancer has been increasing since the 1950s, with the risk being much higher than the national average for those living in the Eastern Cape, particularly in rural areas of the former Transkei "homeland". Since the mid-1980s, the incidence of oesophageal cancer has decreased, as shown by a declining proportion of oesophageal cancers over time in the National Cancer Registry. The reasons for these secular trends remain uncertain.

There are marked differences among population groups, with the highest incidence rates in the African population. The National Cancer Registry recorded a shift from oesophageal cancer as the leading cancer among African males in 1995, to prostate cancer in 1996 to 1999. Oesophageal cancer then became the second leading cancer in African males (with an age-standardized incidence rate [ASR] of 14.1/100 000), who had a lifetime risk of developing oesophageal cancer of 1 in 59.0 in 1999. Among African females (with an ASR 7.0/100 000), oesophageal cancer was the third leading cancer (after cervical and breast cancers), and the lifetime risk was 1 in 113 (Mqoqi et al., 2004).

Because of its poor prognosis, oesophageal cancer contributes significantly to cancer mortality. In South Africa, it was the second leading cause of cancer deaths in males (17.2 percent of all male cancer deaths) and the fourth leading cause in females (10 percent). It was the leading cause of male cancer deaths in the African population (age-standardized mortality rate 43.5/100 000) and the second leading cause - after cervical cancer - in African females (16.5/100 000), with relatively young age groups affected and rates increasing steadily from the 35 to 44 years age group (Bradshaw et al., 2003).

The main risk factors for oesophageal cancer are tobacco use and alcohol consumption, and the joint effect of these is multiplicative (Tuyns, Pequinot and Jensen, 1979; Day, 1984; IARC, 1988). Other possible risk factors include poor socio-economic conditions, poor nutritional intake and a diet lacking in vitamins A and C, riboflavin, nicotinic acid, magnesium and zinc (Cook-Mozaffari et al., 1979; Van Rensburg, 1981). Contamination of maize with Fusarium verticillioides (previously known as Fusarium moniliforme) and the consequent ingestion of mycotoxins (possibly fumonisins) produced by this fungus may also play a role.

Breast cancer

In South Africa from 1984 to 1986, breast cancer age-standardized mortality rates were highest among coloured urban females (26.4/100 000) (Figure 28), followed closely by whites (26.0/100 000) and then urban Indians (14.6/100 000); black urban females had the lowest rates (9.6/100 000) (Bradshaw et al., 1995). In 2000, the age-standardized mortality rates appeared to have increased since 1984 to 1986, although the 2000 national rates were not available by urban/rural residence, and the rate in urban females was probably higher than the national average. As before, the age-standardized death rates for white females were almost threefold those of blacks: whites had the highest rates (33.0/100 000), followed closely by coloureds (31.0/100 000) and Indians (27.4/100 000); black females had the lowest rates (12.1/100 000) (Bradshaw et al., 2003).

However, the age-specific death rates indicate that black females aged 35 to 44 years have very similar rates to those in the corresponding white, coloured and Indian population groups, and it is only in the older age groups that black females have far lower rates. This pattern is also evident in terms of incidence.

Age-standardized mortality rates for females from breast cancer, 1985 and 2000

Note: 2000 figures are adjusted for misclassification and underregistration of deaths, but not 1985 figures. 1985 figures for blacks are for urban blacks only.

Sources: 1985 - Bradshaw et al., 1995; 2000 - Bradshaw et al., 2003.

Black females consistently also had the lowest breast cancer incidence rates. In 1999, the national ASR in blacks was 18.4 per 100 000, compared with 76.5 for coloured and white females (Mqoqi et al., 2004). The differences in rates are more pronounced in older age groups; in the younger age groups, the incidence rates in black females are closer to those of white and coloured females.

The risk of breast cancer is clearly associated with high socio-economic status, and women with higher education or income are at higher risk (Parkin et al., 2003). These differences may be because of differences in the distribution of risk factors among social classes, such as reproductive factors and other known risk factors for breast cancer, including alcohol, diet, smoking, body weight, physical activity and genetic factors. Increased body weight has been found to increase the risk of breast cancer, whereas physical activity has been found to be beneficial in reducing the breast cancer risk at all ages.

Colorectal cancer

Cancers of the colon and rectum are the second most common malignancy in affluent societies, but are rarer in developing countries. In South Africa, colorectal cancer was the sixth leading cancer among males (5.3 percent) and the fifth among females (6.6 percent) in terms of deaths (Bradshaw et al., 2003). In 1999, colorectal cancer comprised 3.7 percent of all cancer cases in males and 3.4 percent in females, ranking third and fifth in females and males, respectively. The ASR for colorectal cancer in women was 6.6 per 100 000, while males had a higher rate of 9.7 (Mqoqi et al., 2004).

Colorectal cancer was the second leading cause of cancer deaths in the South African white population. The age-standardized death rate was more than five times greater in this population (21.1/100 000) than in the black population (4.1/100 000). Colorectal cancer incidence rates were also highest among white males and females. In 1999, colorectal cancer was the second leading cancer in white males and females in terms of incidence. Coloured males and females had the second highest rates, followed by Indian males and females, with the lowest rates reported in black males and females. The rates in white males (ASR of 25.4/100 000) are more than eight times those found in black males (3.0/100 000), while those in white females (17.5/100 000) are about seven times those in black females (2.3/100 000) (Mqoqi et al., 2004).

Age-specific incidence rates by population group suggest an increased risk in younger black South Africans, probably caused by changing lifestyle and diet, resulting in a reduction in the incidence gap observed in elderly South Africans. Although there is an almost tenfold difference in incidence among the older age groups (75 years and over), at younger ages the incidence rates among the black, white and coloured population groups are almost the same.

A diet high in energy (calories), rich in animal fat and poor in vegetables, fruit and fibre is associated with increased risk. Smoking, meat and alcohol consumption are known risk factors, while consumption of fruit and vegetables and physical activity are known to be protective. Hence, the importance of a healthy lifestyle cannot be overemphasized in the prevention of cancers.

[11] These population group classifications reflect self-reporting according to groups defined by the Population Registration Act of 1950. This classification highlights issues that reflect the effects of historical disparities, and the authors do not subscribe to it for any other purpose. The terms "black" and "African" are used interchangeably.
[12] People living in a city all their lives have spent 100 percent in an urban area. Alternatively, people living in rural areas for most of their lives (50 years) with five years in the city have spent only 10 percent of their lives in an urban area

Previous Page Top of Page Next Page